1
2
3
4
5
6
An Internal ALD-Based High Voltage Divider and
Signal Circuit for MCP-based Photodetectors
Bernhard Adamsb , Andrey Elagina , Jeffrey W. Elamb , Henry J. Frischa ,
Jean-Francois Genata,1 , Joseph S. Gregarb , Anil U. Maneb , Michael J. Minotc ,
Richard Northropa , Razib Obaida,2 , Eric Oberlaa , Alexander Vostrikova ,
Matthew Wetsteina
a Enrico
Fermi Institute, University of Chicago
National Laboratory
c Minotech Engineering, Inc and Incom, Inc
7
b Argonne
8
9
10
Abstract
We describe a pin-less design for the high voltage (HV) resistive divider
of the all-glass LAPPDTM 8”-square thin photodetector module. The divider,
which distributes high voltage applied to the photocathode to the two microchannel plates (MCP’s) that constitute the amplification stage, is comprised of
the two MCP’s and three glass mechanical spacers, each of which is coated with
a resistive layer using Atomic Layer Deposition (ALD). The three glass grid
spacers and the two MCP’s form a continuous resistive path between cathode
and anode, with the voltages across the MCP’s and the spacers determined by
the resistance of each. High voltage is applied on an external tab on the top
glass window that connects to the photocathode through the metal seal. The
DC ground is supplied by microstrips on the bottom glass plate that form the
high-bandwidth anode. The microstrips exit the package through the glassfrit seal of the anode base-plate and the package sidewall. The divider is thus
completely internal, with no HV pins penetrating the low-profile flat glass package. Measurements of the performance of the divider are presented for the 8”square MCP and spacer package in a custom test fixture and for an assembled
externally-pumped LAPPDTM prototype with an aluminum photocathode.
1 Present
2 Present
address, Universités Pierre et Marie Curie; Paris France
address, Physics Dept., Univ. of Conn., Storrs, CT 06269
Preprint submitted to NIM
December 26, 2014
11
Contents
12
1 Introduction
3
13
2 The HV Resistive Distribution Chain
5
14
2.1
Resistive Components . . . . . . . . . . . . . . . . . . . . . . . .
6
15
2.2
DC Current Path . . . . . . . . . . . . . . . . . . . . . . . . . . .
8
16
2.3
The Signal Path for Fast-Risetime Pulses . . . . . . . . . . . . .
10
17
2.3.1
Normal Anode Configuration . . . . . . . . . . . . . . . .
11
18
2.3.2
The ‘Inside-Out’ Anode . . . . . . . . . . . . . . . . . . .
12
19
20
21
3 Implementation in the LAPPDTM Package
3.1
Mechanical Dimensions
. . . . . . . . . . . . . . . . . . . . . . .
4 Performance of the HV Divider
14
16
Test Facilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17
23
4.1.1
The Spacer Station (ISS) . . . . . . . . . . . . . . . . . .
17
24
4.1.2
The Demountable Tile . . . . . . . . . . . . . . . . . . . .
17
22
4.1
14
25
4.2
Current-vs-Voltage (I-V) Curves . . . . . . . . . . . . . . . . . .
20
26
4.3
Thermal Coefficient
. . . . . . . . . . . . . . . . . . . . . . . . .
20
27
4.4
Pressure Dependence . . . . . . . . . . . . . . . . . . . . . . . . .
21
28
4.5
Long-term Stability . . . . . . . . . . . . . . . . . . . . . . . . . .
22
29
5 Conclusions
23
30
6 Acknowledgments
24
2
31
1. Introduction
32
The LAPPD Collaboration [1] was formed to develop large-area photodetec-
33
tors based on micro-channel plates [2] for precise time-of-flight measurements
34
in large-scale detectors for High Energy Physics. In a number of applications
35
the ability to detect the arrival of multiple photons in a large-area plane, each
36
measured with a time resolution measured in pico-seconds (1 psec = 10−12 sec)
37
and a space resolution measured in millimeters, enables a 3D pattern recon-
38
struction using the 2D position of arrival of each photon and the transit time
39
to determine the remaining orthogonal 3rd component [3]. Additionally, a psec-
40
level measurement of the time coordinate allows a qualitative change in the
41
photodetector area needed to provide a given level of coverage for large diffuse
42
sources such as the large water-Cherenkov or scintillation counters used in par-
43
ticle and nuclear physics, as the addition of the transit drift time of the photon
44
transforms a 2-dimensional phase space into a much larger 3-dimensional space.
45
Optical systems can take advantage of measuring both the position and the time
46
of arrival to minimize expensive photocathode area while still preserving space
47
resolution3 .
48
The amplification in conventional discrete-dynode photomultipliers (PMT’s)
49
and microchannel-based photomultipliers (MCP-PMT’s) is done by electron
50
multiplication, with the electrons accelerated between collisions with a dyn-
51
ode (in a PMT) or the wall of a capillary channel (in an MCP-PMT) to provide
52
energy for secondary electron emission. The multiplication is consequently ex-
53
ponential, and depends on the secondary-emission yield (SEY) of each collision
54
and the number of collisions (equal to the number of dynodes in a PMT). The
55
SEY in turns depends on the accelerating voltage between collisions, which in
56
a PMT is typically set by a resistive voltage divider that supplies each dynode
57
from a high voltage (HV) source at the input to the divider. The principle is the
3 One
can think of this as similar to rotations between transverse and longitudinal emit-
tances in accelerators.
3
58
same in an MCP-PMT, but rather than a divider made of discrete resistive ele-
59
ments, the voltages between collisions are set by a DC current flowing through a
60
resistive layer on the walls of the capillaries. In an MCP-PMT with the typical
61
pair of capillary plates as an amplification section, voltage is usually applied
62
to the top and bottom of each plate by deposition of a conducting electrode
63
surface.
64
The LAPPDTM (Large Area Picosecond Photodetector) is an MCP-based
65
photodetector, capable of imaging at high spatial and temporal resolution, in an
66
ultra-high vacuum (UHV) hermetic package with an active area of 400 square
67
centimeters. The modular detector design is based on a hermetic all-glass vac-
68
uum tube package consisting of a thin window on which a bialkali photocathode
69
has been deposited, two 8”-square micro-channel plates (MCP), and an anode
70
implemented as high-bandwidth RF silver micro-striplines [4]. Amplification is
71
provided by a pair of micro-channel glass substrates [5] coated by Atomic Layer
72
Deposition (ALD) [6], with typical gains of more than 107 .
73
The HV distribution for the LAPPDTM can be implemented in a number of
74
ways, depending on the application and possibly also on parameters of the pro-
75
duction process. In the design described here, the HV distribution and the high
76
frequency signal path are both integral parts of an economical mechanical struc-
77
ture. The hermetic package has only 8 parts, all made of commonly-available
78
sheet glass. The only additional element is a simple getter assembly formed
79
from a strip of getter material and glass bead supports that is situated around
80
the circumference of the channel plate stack assembly. An attractive feature
81
for robustness and fabrication yield is there are no pins penetrating the pack-
82
age; both the HV and the anode signals are brought in through metal layers
83
deposited on the glass prior to assembly. The planar anode signal plane has an
84
analog bandwidth of 1.6 GHz (3 db loss) [4] in this implementation.
85
The organization of the paper is as follows. Section 2 describes the resistive
86
HV distribution chain, with the resistive elements and DC current path de-
87
scribed in Sections 2.1 and 2.2, respectively. The high-speed (radio-frequency,
88
RF) signal circuit, which is electrically and mechanically integrated with the
4
89
HV circuit, is discussed in Section 2.3 for the usual configuration with the an-
90
ode pattern on the inside surface of the vacuum volume (Section 2.3.1), and a
91
capacitively-coupled configuration with pads or microstrips on the outside sur-
92
face (Section 2.3.2). The integration of the HV divider into the LAPPDTM her-
93
metic package is described in Section 3, where the physical dimensions, ALD
94
coating parameters, and electrical values are also summarized.
95
Measurements of the performance of the HV divider are presented in Sec-
96
tion 4. A summary of the test facilities for the HV chain is given in Sec-
97
tion 4.1. Section 4.1.1 describes a custom test facility, the Spacer Station (ISS),
98
used to make DC measurements of current and voltage. A second test facility,
99
the ‘Demountable Tile’, consisting of an LAPPDTM glass-body tile mechanical
100
assembly, but with an O-ring window seal and a permanent connection to a
101
vacuum pump rather than being a hermetically sealed tube, is used to make
102
measurements of fast pulses, including time and space resolutions, as described
103
in Section 4.1.2. Measurements of I-V curves and the dependence of the divider
104
resistance on temperature, pressure, and time are presented in Sections 4.2-4.5.
105
The paper concludes with a summary in Section 5.
106
2. The HV Resistive Distribution Chain
107
The hermetic package, made of readily-available borosilicate glass [7], con-
108
sists of a top window, sidewall, and anode bottom plate. The sidewall is her-
109
metically sealed to the anode plate with glass frit to form a robust base unit.
110
The top window is sealed with indium to the base in vacuum after the photo-
111
cathode is deposited. HV is applied outside the vacuum volume on a window
112
tab that makes contact with the metalized border of the window and the photo-
113
cathode. Resistive grid spacers coated by ALD support the pressure on the top
114
window and bottom anode plate. The three spacers and the two MCP plates,
115
together with the (DC) grounded anode plane, form the resistive HV divider
116
that determines the voltages across the MCP’s and the three gaps. The silver
117
anode strips [8] extend beyond the sidewall bottom seal to complete the electri-
5
Figure 1: The mechanical construction of the LAPPDTM module. All components are glass. The hermetic volume is formed by the top window (1), the
side-wall (5), and the bottom plate (7). The photocathode is deposited on the
inside of the window, and makes contact with the external HV input through a
metalized border (4). The grid spacers (2) and the MCP’s (3) are coated with
ALD. The bottom plate (7) has silk-screened silver anode 50Ω microstrips fired
on its surface (6) that penetrate a frit seal under the sidewall and connect to
the digitizing electronics.
118
cal circuits for both signals and DC current. Atmospheric pressure transmitted
119
through the top window and bottom anode plane compresses the ’stack’ of grid
120
spacers and microchannel plates into a rigid thin package.
121
122
A diagram of the layers of the HV divider is shown in Figure 1.
2.1. Resistive Components
123
Amplification in the tile is provided by a pair of microchannel-plates with
124
20 µm -diameter pores, arranged in a chevron configuration. The pores have a
125
length-to-diameter ratio (L/D) of 60, making the thickness of each plate 1.2 mm.
126
The plates are coated with a resistive layer by ALD with a target resistance of
127
between 10 and 40 MΩ [9]. The plates are chosen within a pair to be matched
128
in resistance within ±10%.
129
To provide structural rigidity and the ability to support atmospheric pressure
130
on the top and bottom of the package, the gaps in the vacuum volume from the
131
cathode to the first MCP, from the first to the second MCP, and from the
6
Figure 2: The equivalent electrical circuit of the tile assembly.The upper diagram shows the DC circuit. The HV voltage divider is implemented by resistive
ALD coatings of the grid spacers and the MCP’s. (see Figure 1).
132
second MCP to the anode, are formed by spacers consisting of thin flat glass
133
grids. The HV resistive divider chain is completed by coating the grid spacers
134
with a resistive layer, also by ALD, as shown in Fig. 2.
135
The parameters of the layers that make up the tile are given in Table 1 [10].
136
A 2-mm grid spacer, shown in Figure 4, sets the height and voltage across the
137
photocathode-MCP gap and the gap between the two MCP’s. A 6.5 -mm grid
138
spacer similarly sets the height and voltage across the gap between the bottom of
139
the second MCP and the anode plane. The grid spacers are aligned to minimize
140
the occluded area and to transfer the force of atmospheric pressure in columnar
141
fashion between the top and bottom plates.
142
A 2.3 cm by 0.076 cm cut-away indentation in each crossbar in one direction
143
in the bottom spacer allows the vacuum volumes defined by the spacer grid to
144
communicate with each other for pumping, as shown in Figure 3.
145
We have made tests with a Nichrome coating on the grid spacers and without.
146
Because the MCP’s are metalized with Nichrome top and bottom there is no
147
need to metalize the grid spacers at those interfaces. However metalization of
7
148
the top surface of the top spacer can provide a convenient method of distributing
149
HV and current across the face of the (resistive) photocathode. The silver anode
150
microstrips distribute the DC ground potential under the bottom spacer, which
151
cannot be metalized without shorting the strips.
152
2.2. DC Current Path
153
The DC electrical circuit of the photodetector is shown in Figure 2. High
154
voltage is applied to a Nichrome border on the underside of the window via a
155
tab that extends beyond the sidewall. The Nichrome makes contact with the
156
top grid spacer to supply current to the resistive divider through the spacers
157
and MCP’s.
Figure 3: A partial assembly showing the ‘bottom’ spacer grid, the getter strip
and its supporting glass beads, and the sidewall as they would be in the assembled tile base. The 2.3 cm by 0.076 indentations that allow the separate vacuum
volumes in the tile to communicate with each other are visible as channels running from left to right on the top edges of the spacer.
8
158
Figure 5 shows the window, including the external tabs and Nichrome bor-
159
der that supply the input HV to the photocathode and the divider. The HV
160
is applied to the underside of two of the tabs, and connects to the cathode
161
through the metalization for the seal. The four metalized lines that extend into
162
the cathode provide HV distribution to the cathode plane, complementing the
163
distribution from the top spacer.
164
The internal five layers, consisting of three grid spacers and two MCP’s, form
165
a resistive voltage divider [11] that passes a nominal 100 µA divider current from
166
the cathode to the anode. Table 2 gives the nominal resistance and voltage drops
167
across the divider.
168
The final step in the DC return path to HV ground is at each end of the
169
anode microstrips, where individual 10KΩ resistors terminate each strip to the
170
copper substrate sheet that forms the signal ground-plane of the microstrips
Layer
Thickness
Material
Resistive Coating
Metalization
(side)
Window
2.75 mm
B33
Photocathode
Nichrome border
(bottom)
(bottom)
Grid Spacer 1
2.0 mm
B33
ALD-GS
None
MCP 1
1.2 mm
Micropore
ALD-MCP
Nichrome (both)
Grid Spacer 2
2.0 mm
B33
ALD-GS
None
MCP 2
1.2 mm
Micropore
ALD-MCP
Nichrome (both)
Grid Spacer 3
6.5 mm
B33
ALD-GS
None
Anode
2.75 mm
B33
N/A
Silver strips (top)
Table 1: A tabulation of the layers in the tile assembly, starting with the top
window and ending with the bottom anode layer. Mechanical stability against
atmospheric pressure is provided by a glass grid spacer in each of the three gaps,
as listed. All components except the microchannel plates are made from B33
borosilicate glass plate [7].
9
171
and the HV DC return, as shown in Fig. 2.
172
2.3. The Signal Path for Fast-Risetime Pulses
173
The pulses from photons incident on the photocathode have sub-nsec rise-
174
times, with typical bandwidths for the present 20-micron pores in the 1-2 GHz
175
range. The DC HV path and the path of the fast signal pulses share the same
176
HV distribution and ground. For the signal path, however, there are two con-
177
figurations we have considered, a ‘normal’ one in which the charge of the fast
178
signals is collected and recorded from the anode microstrips inside the vacuum
179
volume, and an ‘inside-out’ geometry in which the anode is a thin metal layer
180
that serves as a DC ground, but is thin enough so that the fast signal pulses
181
capacitively couple through to strips or pads on the external side of the bottom
182
plate. These are described in turn below.
Figure 4: Two of the glass grid spacers after the resistive ALD coating has
been applied. The grid spacers are cut from easily available B33 glass [7].
Mechanically, the spacers serve to transmit the compressive force of atmospheric
pressure from the top to the bottom of the tube; electrically the resistive coating
provides the voltage drop across each gap. Metalization of the top surface of
the top spacer can serve to provide distribution of HV across the (resistive)
photocathode.
10
183
2.3.1. Normal Anode Configuration
184
The signal path maintains the 50Ω impedance of the strip onto the printed
185
circuit card containing the waveform sampling 6-channel PSEC4 ASICs [12]
186
that digitize the signals. The thickness of the anode glass plate and the width
187
of the strips determine the impedance of the microstrip lines [4, 13]. While
188
the choice of impedance depends on a number of considerations, we chose a
189
50Ω impedance for convenience in testing to match cabling and oscilloscope
190
impedances. The choice of the anode plate thickness is application specific,
191
with impedance, strip parameters, occupancy, weight, cost, amount of inactive
Figure 5: The glass window, showing the Nichrome border that is the ‘tie layer’
supporting the indium seal to the glass sidewall. The external tabs and the
metalized border also supply the input HV to the cathode and resistive divider.
The view is from the air-side (top); the Nichrome metalization for the seal
and the HV distribution is visible on the vacuum-side (underneath) along the
perimeter and on the four lands extending into the interior of the cathode.
11
Layer
Resistance
Voltage Drop
Grid Spacer 1
2 MΩ
200 V
MCP 1
10 MΩ
1000 V
Grid Spacer 2
2 MΩ
200 V
MCP 2
10 MΩ
1000 V
Grid Spacer 3
6.5 MΩ
650 V
Total
28.5 MΩ
3050 V
Table 2: The target resistances and nominal voltage drops in the internal HV
divider. The divider current is set at 100 µA, corresponding to a limit of approximately 1 µA of current draw from signal pulses [11].
192
material in the path of charged particles being some of the considerations. After
193
the plate thickness is chosen the width of the anode strips can be set so to
194
determine the characteristic impedance. A further parameter is the gap between
195
strips, which is vulnerable to charging in high-rate applications if electric field
196
lines terminate on the dielectric.
197
Each 50Ω stripline is impedance-matched to the PC cards that digitize the
198
signals and that mate with the edge of the end tiles of a tile row, as shown in
199
Fig. 6. The DC return for the divider current through the 10KΩ resistor has
200
a much higher impedance than the 50Ω signal path. For flexibility in testing,
201
an easily removable connection was made using low profile Electro-Magnetic
202
Interference (EMI) shielding gaskets [14]. Each EMI gasket strip was cut at
203
the pitch interval into separate RF fingers and mounted on a 1/2” G10 bar to
204
match the anode spacing. Figure 6 shows the connections between the strips on
205
the tile and the strips on the PC card used for testing the tiles.
206
2.3.2. The ‘Inside-Out’ Anode
207
There are a number of applications, such as large collider detectors in HEP
208
and some medical imaging cameras, that would have many particles arriving in a
209
narrow time window and consequently high occupancies in a strip-line readout.
12
Figure 6: A picture showing the interface between the tile anodes and the
readout system at the end of a row of tiles. The DC path to HV ground at the
end of the microstrips is through individual 10KΩ resistors on each strip to HV
ground on the analog card that holds the waveform sampling ASICs. To allow
easy disconnection for testing, a bar of FR4 with RF-fingers is used to bridge
the gap between each of the 30 strips on the tile to the corresponding strip on
the PC card.
13
210
These applications would benefit from a 2-dimensional ‘pad’ readout rather than
211
a 1-dimensional stripline geometry.
212
The fast risetime of the signal pulses from the MCP amplification chain
213
provides a mechanism to extract the signals from the vacuum volume by ca-
214
pacitively coupling through the glass anode bottom plane. In this ‘inside-out’
215
implementation, a uniform metal layer thinner than a skin-depth at frequencies
216
characteristic of the MCP risetime is deposited on the inner (vacuum-side) side
217
of the anode bottom glass plate, replacing the microstrips. This layer provides
218
the ground return for the DC HV current. The outer (air-side) side of the anode
219
plate supports the readout pattern deposited by metalization, typically in pads
220
or strips. This layer can then be connected to the electronics and additional
221
grounds as necessary.
222
3. Implementation in the LAPPDTM Package
223
The LAPPDTM design integrates the mechanical structure with both the
224
high frequency signal path and the HV DC current path into a single economical
225
structure. The advantage of the integrated design is the construction is planar,
226
with all electrical connections lying in the same plane as the thin form-factor
227
detector module itself. This allows a ‘tiling’ of large areas with a higher filling
228
factor than would be possible with individual modules with edge-readout. We
229
give details of the implementation below.
230
3.1. Mechanical Dimensions
231
The tile assembly is a flat glass package with transverse dimensions of 229.1
232
mm in the strip direction by 220 mm transverse. The active area is 203 mm
233
by 203 mm (8” by 8”). The current prototype grid spacers occlude 16% of the
234
active area; this will be reduced in future designs after more experience with
235
this first-generation simple implementation.
236
A mockup of an assembled glass module, made with prototype glass parts
237
but without a photocathode or vacuum seal, is shown in Figure 7. Multi-
238
ple modules can be integrated into a ‘supermodule’, sharing the RF-strip-line
14
239
readout mounted on the ‘Tray’ that also supports the electronics, as shown in
240
Figure 8 [4].
Figure 7: An 8”-square (20 cm-by-20 cm) LAPPDTM module (‘tile’). The tiles
are hermetically sealed vacuum tubes that provide the photocathode and MCPbased amplification chain. The upper layer of the micro-strip RF transmission
lines that form the anode are on the internal surface of the bottom plate of
the tile; the ground plane of the transmission line is the upper surface of the
supporting ‘Tray’, shown in Figure 8. The thin format and modular nature of
the photodetector allow application-specific optimization.
241
The tile is made from glass layers stacked to form a rigid structure when
242
compressed by atmospheric pressure. The vacuum volume is enclosed by the
243
2.75 mm-thick top window, a standard plate glass thickness. The photocathode
244
is on the bottom surface of the window, facing the top of the upper MCP (see
245
Fig. 1). The anode plate with silver microstrips on its top surface, serves as the
246
bottom of the vacuum volume, and is also 2.75 mm-thick. A rectangular border
247
of 5.08 mm (0.200”) width, cut from a standard thickness glass plate, forms the
15
Figure 8: A ‘supermodule’ consisting of twelve 8”-square LAPPDTM modules
(‘tiles’) mounted on a ‘tray’ that provides the ground plane for the anode microstrips and the support for the digitizing electronics at each end. Each row of 4
tiles shares the same anode strips in series, as described in Ref. [4].
248
sidewall of the volume.
249
The sidewall is sealed to the anode plate over the microstrips with a glass
250
frit seal. The metal microstrips extend beyond the sidewall so that contact can
251
be made with neighboring tiles [4] or the digitization printed circuit cards at
252
the end of a tile-row [12].
253
4. Performance of the HV Divider
254
The proposal to use ALD coatings on the grid spacers and MCP’s to deter-
255
mine the amplification and transport voltages had unknown risks at the outset
256
of the LAPPD program. In particular we are relying on the stability of the
257
coatings under a large mechanical pressure. Atmospheric pressure corresponds
258
to almost 1200 lbs on the top and bottom plates. The pressure on the MCP
16
259
surfaces is magnified by the ratio of areas of the grid spacer to the total area,
260
approximately a factor of 10. The surfaces of the MCP’s consist of glass capil-
261
laries with ∼ 65% open area, and have both metal electrode material and ALD
262
secondary-emitting material; the surfaces of the grid spacers have a grid-specific
263
ALD resistive coating. How these interfaces behave under pressure, with tem-
264
perature, and over time is not tractable to calculation; we have consequently
265
embarked on a series of measurements of the stability of the HV divider assem-
266
bly under HV and pressure. Several of the test facilities are described in turn
267
below.
268
4.1. Test Facilities
269
Section 4.1.1 describes a custom test facility, the Spacer Station (ISS) used
270
to make DC measurements of current and voltage. A second test facility, the
271
‘Demountable Tile’, consisting of an actively-pumped glass-body tile mechanical
272
assembly with an O-ring top seal, is used to make measurements of fast pulses
273
at resolutions down to 5 psec, as described in Section 4.1.2.
274
4.1.1. The Spacer Station (ISS)
275
To measure the resistance of the grid spacers and MCP plates versus pres-
276
sure, temperature, and over time, a simple flat square vacuum vessel with a
277
flexible lid consisting of an aluminum foil incapable of withstanding atmospheric
278
pressure was constructed. The window provides the same (atmospheric) pres-
279
sure on the component stack as the glass window of the sealed tile. Figure 9
280
shows the ISS open to air with a single grid spacer inside. The copper plate
281
to the right of the ISS provides HV contact to the layers being measured; the
282
ISS frame is the DC ground. Single, multiple, or a full stack of the component
283
layers (grid spacer and MCP) can be measured using suitable non-conducting
284
spacers above the copper plate.
285
4.1.2. The Demountable Tile
286
In order to test the time and position response of the photodetectors as a
287
system we have constructed a facility [15] of which the key components are a
17
Figure 9: The spacer station (ISS) used to measure resistances of individual
grid spacers, MCP’s, and the complete divider. A grid spacer is shown in the
vacuum volume; the copper plate being held on the right is an electrode that
supplies the HV to the piece being measured. A thin insulating layer separates
the copper HV electrode from the 2.75-mm-thick top window, which is thin
enough that atmospheric pressure deflects it until stopped by the internal stack
of MCP’s and grid spacers, compressing the divider.
288
pulsed Ti:Sapphire laser with a pulse width ∼100 fsec, and a data-acquisition
289
system with 60 channels of 10-15 GS/sec waveform sampling readout, shown
290
in Figure 10. The Demountable Tile is constructed with the anode frit-sealed
291
to the sidewall and the internal HV divider as in the LAPPDTM , but has
292
a removable window sealed with an O-ring so that MCP plates can be easily
293
inserted and removed for rapid testing. The tile consequently has to be actively
294
pumped, and has an air-stable aluminum cathode.
18
295
The HV control system for the Demountable Tile Test Facility allows for
296
measuring voltage and current under computer control [15], and similarly mea-
297
suring the stability of the gain and spatial response over the face of the tile [16].
298
Figure 10: The Demountable Tile test setup, consisting of glass tile of the
LAPPDTM design constructed with the LAPPDTM hermetic package and internal divider, but with a removable window sealed with an O-ring. The tile
consequently has to be actively pumped, and, as a bialkali anode cannot survive
in air, has an aluminum cathode. However the HV divider and signal circuit
are the same as in the sealed tile. The tile is shown sitting on a 4-tile anode
readout ‘Tray’, which is equipped with 60 channels (30 strips on each end) of
10-15 GS/sec waveform sampling PSEC-4 ASICs.
19
299
4.2. Current-vs-Voltage (I-V) Curves
300
One concern about using ALD-coated elements as a HV divider is the possi-
301
ble non-linear or non-reproducible behavior as a function of voltage. Figure 11
302
shows the measured I-V curve using the ISS facility. The behavior is linear, and
303
demonstrates little hysteresis.
Figure 11: The current-vs-voltage (I-V) data for the ALD-based HV divider, as
measured in the ISS test facility. The voltage is measured both ramping up and
ramping down increased (red, squares) vs decreased (blue, circles). The slope
gives the effective resistance of the multi-layer divider.
304
4.3. Thermal Coefficient
305
The temperature dependence of the divider resistance is of concern due to
306
the vulnerability of MCP-based photodetectors to thermal run-away [17]. The
307
divider resistance depends on the thermal coefficients of the ALD-coatings of
308
the grid-spacers and MCP’s. The resistivity of the coatings on the grid spacers
309
is different from that for the MCP’s, as the effective conducting areas are so
310
different. At present the three grid spacers are coated in the same batch, so the
20
311
difference in resistance is determined solely by the difference in the conduction
312
path length of the thicker grid spacer from that of the other two (see Table 1).
313
Figure 12 shows the resistance of the HV divider as a function of temperature,
314
measured in the ISS.
Figure 12: The resistance of the HV divider as a function of temperature. The
measurements between room temperature and the temperature of ice were measured by necessity ‘on the fly’, rather than in thermal equilibrium, a possible
explanation of the apparent offset from the curve.
315
4.4. Pressure Dependence
316
The contacts between the grid spacers and the MCP’s are complex electri-
317
cally and mechanically [18]. The Spacer Station (ISS) was constructed with
318
a thin foil top ‘window’ specifically to measure the contact resistances versus
319
pressure. The pressure is varied by control of the vacuum inside the ISS through
320
a needle valve to the pump. A curve of resistance vs the pressure on the layers
321
of the HV divider (lbs/in2 ) is shown in Figure 13. The drop in resistance of
322
approximately 20% at low pressure is presumably due to changes in contact. A
21
323
small amount of hysteresis is observed; however in an operating evacuated tile
324
the pressure on the stack is stable, with only minor variations due to changes
325
in atmospheric pressure or temperature.
Figure 13: The resistance of the HV divider as a function of the pressure compressing the divider stack. The pressure is varied by using the differential between atmospheric pressure on one side of the thin foil window at the top of
the ISS and the partial vacuum inside the volume. The pressure in the ISS is
controlled by pumping through a needle valve. Note the suppressed zero on the
ordinate.
326
4.5. Long-term Stability
327
The large pressure (> 1000 lbs) on the HV divider transmitted through the
328
thin glass top window and bottom plate is also cause for concern about the
329
long-term stability of the electrical contacts between the layers that constitute
330
the divider. The mechanical and electrical stability of the interface are subject
331
to changes due to pressure, electro-chemistry, and temperature, as well as a
332
possible slow creep of a component or layer under the pressure. Over the course
22
333
of 11 months under vacuum, a stack of two MCP’s and 3 grid spacers the
334
Demountable Tile has remained stable in resistance, gain, and uniformity, giving
335
us confidence in the mechanical design.
336
5. Conclusions
337
The use of ALD for functionalizing glass capillaries and glass grid spacers
338
enables a modular compact design of a thin planar photodetector with an in-
339
ternal HV divider. The LAPPDTM module is simple, consisting of eight parts,
340
all made of glass, and a simple getter assembly. The amplification section con-
341
sists of two 203 mm-by-203 mm MCP’s. The HV divider is formed by applying
342
resistive coatings to the grid spacer in each of the three gaps: between the pho-
343
tocathode and MCP1, between MCP1 and MCP2, and between MCP2 and the
344
anode microstrip transmission lines.
345
The sealed module sits flat on a ground plane that is shared with neighboring
346
modules; there are no connecting pins on the back side. The HV is applied to an
347
external metalized tab that connects to the metal top seal between the window
348
and the glass sidewall. The DC ground is provided by the anode ground plane
349
of microstrips, which are deposited on the top surface of the glass bottom plate
350
of the hermetic package, and extend through the seal to outside the vacuum
351
volume. The anode microstrips can be connected in series to form a continuous
352
anode up to 90 cm long, with the DC ground return and ASIC readout on the
353
ends of the outer-most modules [4].
354
Measurements of current vs voltage in the HV divider have been made in
355
a custom test setup and in the module glass body on a 90-cm anode. The
356
measurements in the test setup have been made as a function of the pressure on
357
the divider, controlled by varying the vacuum in the module. Measurements of
358
the temperature dependence and stability are also presented. An 11-month test
359
under vacuum gives us confidence in the long-term stability of this relatively
360
simple integrated electro-mechanical design.
23
361
6. Acknowledgments
362
We thank our colleagues in the Large Area Psec Photodetector (LAPPD)
363
Collaboration for their contributions and support. Special thanks are due to A.
364
O’Mahony (Incom) and Neal Sullivan (Arradiance) for their essential work on
365
ALD coatings for the components of the internal stack. Thanks are due to R.
366
G. Wagner (ANL) for technical support; D. Walters, and J. Williams (ANL) for
367
metalization of the windows and vacuum expertise; and H. Wen and H. Gibson
368
(ANL) for laser and electronics support at the APS testing lab. We are deeply
369
grateful to E. Hahn (Fermilab) for meticulous metalization of the MCP’s, and P.
370
Murat (Fermilab) for support. R. Metz (UC) provided expert machining of test
371
setups and glass parts. M. Heintz (UC) supplied crucial technical and computer
372
systems support. We thank Q. Guo (UC), Chian Liu (ANL) and H. Clausing
373
(H. Clausing, Inc) for expert advice and large amounts of time teaching us about
374
cleaning and metalization of glass. We also thank our superb glass vendors P.
375
Jaynes (CatI Glass, Inc) and P. Seegebrecht (Webcorr) for metalization and
376
glass
377
The activities at Argonne National Laboratory were supported by the U. S.
378
Department of Energy, Office of Science, Office of Basic Energy Sciences and
379
Office of High Energy Physics under contract DE-AC02-06CH11357, and at
380
the University of Chicago by the Department of Energy under DE-SC-0008172,
381
the National Science Foundation under grant PHY-1066014, and the Driskill
382
Foundation.
24
383
References
384
[1] The original LAPPD institutions include ANL, Arradiance Inc., the
385
Univ. of Chicago, Fermilab, the Univ. of Hawaii, Muons,Inc, SLAC,
386
SSL/UCB, and Synkera Corporation. More detail can be found at
387
http://psec.uchicago.edu/.
388
389
[2] J.L. Wiza, Micro-channel Plate Detectors. Nucl. Instr. Meth. 162, p567;
1979
390
[3] E. Oberla, Ph.D thesis, in preparation. We thank H. Nicholson for the
391
original suggestion of using large-area planar detectors for water Cherenkov
392
neutrino detectors, and for the OTPC name.
393
[4] H. Grabas, R. Obaid, E. Oberla, H. Frisch, J.-F. Genat, R. Northrop, F.
394
Tang,D. McGinnis, B. Adams, M. Wetstein, RF Strip-Line Anodes for Psec
395
Large-Area MCP-based Photodetectors. Nucl. Instr. Meth. A71, p124; 2013
396
[5] The glass capillary substrates are produced by Incom Inc. Charlton Mass.
397
398
See http://www.incomusa.com/.
[6] S. M. George, Atomic Layer Deposition: An Overview; Chemical Reviews
399
2010, 110, (1), 111-131;
400
J. W. Elam, D. Routkevitch, and S. M. George, Properties of ZnO/Al2O3
401
Alloy Films Grown Using Atomic Layer Deposition Techniques; Journal of
402
the Electrochemical Society 2003, 150, (6), G339-G347;
403
D. R. Beaulieu, D. Gorelikov, H. Klotzsch, P. de Rouffignac, K. Saadat-
404
mand, K. Stenton, N. Sullivan, and A. S. Tremsin, Plastic microchannel
405
plates with nano-engineered films; Nucl. Instr. Meth. A633, S59-S61 (2011);
406
O.H.W. Siegmund, J.B. McPhate, S.R. Jelinsky, J.V. Vallerga, A.S. Trem-
407
sin, R. Hemphill, H.J. Frisch, R.G. Wagner, J. Elam, and A. Mane,
408
Development of Large Area Photon Counting Detectors Optimized for
409
Cherenkov Light Imaging with High Temporal and sub-mm Spatial Reso-
410
lution; NSS/MIC 2011 IEEE, p2063 (Oct., 2011).
25
411
[7] http://psec.uchicago.edu/glass/borofloat 33 e.pdf#page=28; The dielec-
412
tric constant is 4.6 and the loss tangent is 37 × 10−4 , both measured at
413
25C and 1 MHz.
414
[8] Ferro Corp., 251 Wylie Ave., Washington PA 15301. We are grateful to E.
415
Axtel for extensive assistance with selecting the silver ink for silk-screening
416
the anode microstrips.
417
[9] A. U. Mane, W. M. Tong, A. D. Brodie, M. A. McCord, and J. W. Elam,
418
Atomic Layer Deposition of Nanostructured Tunable Resistance Coatings:
419
Growth, Characterization, and Electrical Properties, ECS Transactions, 64
420
(9) 3-14 (2014);
421
A. U. Mane and J. W. Elam, Atomic Layer Deposition of W:Al2O3
422
Nanocomposite Films with Tunable Resistivity, Chem. Vap. Deposition, 19,
423
186-193, (2013)
424
[10] The tabulated values are those in current use. They can be changed to
425
optimize the voltages across the respective gaps for specific performance
426
goals. The total thickness depends on the height of the sidewall, which also
427
can be optimized for different applications.
428
429
[11] Photomultiplier Tubes: Principles and Applications, S-O Flyckt and C.
Marmonier; Photonis Brive, France, Sept. 2002
430
[12] E. Oberla, H. Grabas, J.-F. Genat, H. Frisch, K. Nishimura, G. S.
431
Varner, A 15 GSa/s, 1.5 GHz Bandwidth Waveform Digitizing ASIC,
432
Nucl.Instrum.Meth. A735 (2014) 452-461.
433
[13] For an excellent discussion of transmission lines, see E. Bogatin, Signal
434
Integrity Simplified, Prentice Hall Professional Technical Reference, 2004.
435
Chapter 10 is especially clear on cross-talk in striplines.
436
437
[14] Leader Tech part number 12-60LPAH-BD-16; Laird Technologies, Item 97115
26
438
[15] B. Adams, M. Chollet, A. Elagin, R. Obaid, E. Oberla, A. Vostrikov,
439
P.Webster, M. Wetstein; A Test-Facility for Large-area Microchannel Plate
440
Detector Assemblies Using a Pulsed sub-Picosecond Laser, Rev. Sci. In-
441
strum. 84, 061301 (2013)
442
[16] B. Adams, A. Elagin, M. Wetstein et al.; Measurements of the Gain, Time
443
Resolution, and Spatial Resolution of a 2020 cm2 MCP-based Picosecond
444
Photo-detector, Nucl. Instrum. Meth. A732 (2013)
445
[17] C. W. Carlson and J. P. McFadden Design and Application of Imaging
446
Plasma Instruments, in Measurement Techniques in Space Plasmas: Parti-
447
cles, R. F. Pfaff, J.E. Borovsky, and D. T. Young eds.; American Geophys-
448
ical Union pub. (1998)
449
[18] The measurements presented here are made with MCP’s that have the
450
electrode surface ‘under’ the ALD resistive and emissive coatings. The other
451
option, ‘over’, has the electrode on top of the ALD coatings. Both options
452
have their merits and drawbacks.
27